Work in progress on Surface Moisture modelling in the MDV

Work plan

0. write up some stuff I need to know about MDV surface moisture and its behaviour in this RMD as input to the paper

1. explore soil moisture dataset

   • data cleaning necessary?
   • understand surface moisture and temperature information distributions, determine cut-off value (0°C? -x°C?)
   • spatial configuration of loggers, i.e. think about validation strategy (how to do the spatial CV)

2. data gathering for pre-study: find a cloud free day and get as many useful spatial predictor datasets as possible

3. describe and understand relations between surface moisture and predictors & temperature

   • surface moisture and elevation
   • surface moisture and temperature

4. build model for case study

5. run workflow for the whole temporal and spatial setting

To do’s:

• ask Pierre about salinity and pH-paper progress

Paper relevant info

Introduction

Research Question:

• For Method 1 and 2: How accuarate can surface moisture be modeled for the open soil areas within the Antarctic Dry Valleys?

• actually interesting question: Which factors does the spatio-temporal surface moisture configuration in the MDV depend upon?

What we know about surface moisture in the MDV

Take a look at Overleaf note collection

Questions

• should I only use values above 0 degrees? Find out when water freezes in the MDV and how the relation to salinity would be

• How does RH relate to soil moisture?

• can we detect Soil Moisture directly with RS techniques? or is indirect modelling needed?

Data

Calibration and validation

AWS and iButton spatial and temporal distribution

Already available:

• iButton data

To download:

Antarctica Soil Climate Research Stations

Data can be found here.

Get following Variables:

ID, Year, Hour, Date Air Temp (°C, 1.6m) Solar Radiation (W/m²), 3m Wind Speed (m/s), 3m Wind Direction (azimuth), 3m

Soil Temp (°C) Surface under white rock Soil Temp (°C) Surface under black rock

Soil RH (%) 5cm Soil RH (%) 15cm

MRC Soil Temp °C 0 in MRC Soil Temp °C 3 in 7,6cm

Soil 2cm: ER unitless Real dielectric constant EI unitless Imaginary dielectric constant Temp °C Temperature ER_COR unitless Temperature corrected ER EI_COR unitless Temperature corrected EI WATER wfv Soil moisture SALINITY g NaCl/L Soil salinity SOIL_COND S/m Soil conductivity SOIL_COND_COR S/m Temp corrected soil conductivity WATER_CON_COR S/m Temp corrected conductivity of the water in the soil

• Met station AWS - can’t be used, RH measurement at 3m height quality controlled and official data (until 2019 or 2020) non quality controlled but full record until now

Example: Explorer’s Cove

Could be interesting to get precipitation from there

Potential spatial predictors

Potential spatial predictors

Already available:

• DEM 8 / 30m * TWI * slope
• rock outcrop to use as a mask
  • use soil type map to find out where there is only rock and no soil
• LST 30m
• soil types
• pH model (Pierre: “pH is very much correlated to soil moisture, and more linear than EC (because it’s already a log scale!), but I could also share EC estimates if need be.”)

To acquire

• RS data:
  • SWIR
  • downscaled LST
  • radar
• EC (Pierre)?

Methods

• Statistical Exploration of associations
  • check wind relevance with station data?
• Modelling
• Explainable AI to find out which factors are relevant how
• Analysis of Moisture patterns

Method schematic overview

Discussion

References